The thing about edge

[toolazy]

Quote from dbphoenix:

There's a slightly sarcastic tone here, which I assume is on purpose and is done to lighten things up, but I disagree with one and two. A trading strategy requires predictable outcomes and consistent profits. As long as one keeps it simple, this is not difficult if one is willing to do it manually. Unfortunately, since charting software became available to the retail trader fifteen years ago, the computer backtest has become a default, even though there is a serious disconnect between backtested results and what one finds in real-time trading or even forwardtested sim (one finds after a computer backtest that when he begins trading "for real" that he's never even seen this stuff before). Second, if there's no low winrate, there's no psychological challenge. It's a moot issue.

As to the last, yes, largely because traders don't understand risk in the first place, much less the various types of risk. But if one has a high winrate and a high P:L ratio, money management becomes more of a technical issue than a psychological one: where to enter and with how much, where to scale in or out, where to exit and how. And much of this depends on bar interval, timeframe, and style. But one's pants need not be twisted into a knot over it. It's not that much different from the process of putting together a budget. Granted a lot of youngsters have no idea how to do that either, but it's not what one would call monumentally difficult.

so how will you derive edge? every single anomaly is corrected sooner or later. bigger anomaly around longer, more people will notice and greedier and more violent will be reversal in fortune of this anomaly. On top of that you play against corrupt insiders which is costing you long term.

you have to learn from ichiban guy. he gets short and talk market down and he takes profit.

 

[dbphoenix]

Quote from toolazy:

so how will you derive edge? every single anomaly is corrected sooner or later. bigger anomaly around longer, more people will notice and greedier and more violent will be reversal in fortune of this anomaly. On top of that you play against corrupt insiders which is costing you long term.

you have to learn from ichiban guy. he gets short and talk market down and he takes profit.

The edge is obtained via research and testing (see next post). As for "anomalies", if you mean the imbalances between demand and supply, then they won't be "corrected" so much as they will shift back and forth. This is how the Law of Supply and Demand works (as well as auction markets). If people get greedy and violent about it, so much the better. The more emotional traders get, the more profit for those who don't. As for "corrupt insiders", one can take advantage of their activities simply by following in their footsteps.

As for Icahn, one can profit from this just as he might have. Just follow in his wake. The entry at or about 1450 was clear to anyone tracking the imbalance between demand and supply.

This isn't difficult, but if one is closed-minded about it, it's impossible.

 

[dbphoenix]

Re How To Do It (4.5 pages)

 

[toolazy]

Quote from dbphoenix:

Re How To Do It (4.5 pages)

what doc says not to use indicators, etc. How else will you test 3000 samples of same pattern to give you confidence?

How can you include context in your test?

if you don't have confidence, cant win.

 

[dbphoenix]

Quote from toolazy:

what doc says not to use indicators, etc. How else will you test 3000 samples of same pattern to give you confidence?

(a) How do you suppose traders traded before indicators? The indicators commonly used today weren't available until after 1950, and then only to professionals. Indicators weren't available to retail traders until the very late 90s. How did all those people (like me) manage?

(b) Why do you need 3000 samples? And how and why will 3000 samples give you more confidence than 100? There are after all only three strategies, and one needn't be expert in more than one in order to make money.

How can you include context in your test?

http://www.elitetrader.com/vb/showthread.php?s=&postid=3901369&highlight=stickie#post3901369

if you don't have confidence, cant win.

True. If you're interested in a practical example of process in real time, or what was at the time real time, see Game's journal, Making of a Method.

 

[kut2k2]

Quote from dbphoenix:

There's a slightly sarcastic tone here, which I assume is on purpose and is done to lighten things up

No sarcasm was intended.

Quote from dbphoenix:

... but I disagree with one and two. A trading strategy requires predictable outcomes and consistent profits. As long as one keeps it simple, this is not difficult if one is willing to do it manually. Unfortunately, since charting software became available to the retail trader fifteen years ago, the computer backtest has become a default, even though there is a serious disconnect between backtested results and what one finds in real-time trading or even forwardtested sim (one finds after a computer backtest that when he begins trading "for real" that he's never even seen this stuff before).

When did consistent profits become incompatible with positive expectation?

And do it manually? Wm Eckhardt doesn't trust any strategy tested on fewer than 1800 trades. Try doing that manually, just for the testing phase of each candidate strategy. I'm not as rigid as Eckhardt but I do like to see several hundred trades before I trust test results. As to the alleged unreliability of backtesting, that's due to lazy traders taking shortcuts like assuming there are no commissions. Backtests can be made as realistic as any forward sim if a trader cares to take the time and effort to do it.

Quote from dbphoenix:

Second, if there's no low winrate, there's no psychological challenge. It's a moot issue.

That's a huge IF. A trader can't just wish low winrate away if it's integral to the best winning strategy he can devise. The Turtles did make millions with a low-winrate strategy, so it's doable.

Quote from dbphoenix:

As to the last, yes, largely because traders don't understand risk in the first place, much less the various types of risk. But if one has a high winrate and a high P:L ratio, money management becomes more of a technical issue than a psychological one: where to enter and with how much, where to scale in or out, where to exit and how. And much of this depends on bar interval, timeframe, and style. But one's pants need not be twisted into a knot over it. It's not that much different from the process of putting together a budget. Granted a lot of youngsters have no idea how to do that either, but it's not what one would call monumentally difficult.

If one has a high winrate and a high P:L ratio, then one has a great strategy. Again, that's a huge assumption to make for most traders who do manage to make it into PE territory. Sure the higher your Kelly fraction, the more room for error you have if you don't know exactly where best to place your position size. We should all be so fortunate, but most aren't. So knowledge of how best to estimate one's optimal position size might become critical.

 

[toolazy]

Quote from dbphoenix:

sorry, I probably mixed up with someone else. I thought you wrote that one needs to test pattern over 10 years period and many thousand samples.

This is contradictory with the approach in document as one needs many years to get close to above numbers.

Else I like reading about your approach.

 

[dbphoenix]

When did consistent profits become incompatible with positive expectation?

I didn't say they were. Consistent profits are a necessary component of an edge.

And do it manually? Wm Eckhardt doesn't trust any strategy tested on fewer than 1800 trades. Try doing that manually, just for the testing phase of each candidate strategy. I'm not as rigid as Eckhardt but I do like to see several hundred trades before I trust test results.

That's your choice, and Eckhardt's, but it's not a given.

As to the alleged unreliability of backtesting, that's due to lazy traders taking shortcuts like assuming there are no commissions. Backtests can be made as realistic as any forward sim if a trader cares to take the time and effort to do it.

Depends on how it's conducted. If the trader is looking only at the results with which the computer provides him rather than the charts themselves, then the actual trading is going to be quite a surprise.

Quote from dbphoenix:

Second, if there's no low winrate, there's no psychological challenge. It's a moot issue.

That's a huge IF. A trader can't just wish low winrate away if it's integral to the best winning strategy he can devise. The Turtles did make millions with a low-winrate strategy, so it's doable.

If a low winrate is integral to the best winning strategy that the trader can devise, he's doing it wrong. What anybody else does is not necessarily pertinent.

Quote from dbphoenix:

As to the last, yes, largely because traders don't understand risk in the first place, much less the various types of risk. But if one has a high winrate and a high P:L ratio, money management becomes more of a technical issue than a psychological one: where to enter and with how much, where to scale in or out, where to exit and how. And much of this depends on bar interval, timeframe, and style. But one's pants need not be twisted into a knot over it. It's not that much different from the process of putting together a budget. Granted a lot of youngsters have no idea how to do that either, but it's not what one would call monumentally difficult.

If one has a high winrate and a high P:L ratio, then one has a great strategy. Again, that's a huge assumption to make for most traders who do manage to make it into PE territory. Sure the higher your Kelly fraction, the more room for error you have if you don't know exactly where best to place your position size. We should all be so fortunate, but most aren't. So knowledge of how best to estimate one's optimal position size might become critical.

It might. But all that comes later. First one must understand what it is that he's looking at, then what to look for, then what to do with whatever he's looking for if and when he finds it, without fear. Once he begins profiting from his work, then he can concern himself with position sizing.

Incidentally, what is or is not off-topic is entirely up to you. If I venture off into areas which are of little or no interest, just say so and I'll stop.

 

[dbphoenix]

Quote from toolazy:

sorry, I probably mixed up with someone else. I thought you wrote that one needs to test pattern over 10 years period and many thousand samples.

This is contradictory with the approach in document as one needs many years to get close to above numbers.

There is that going around, but it's not a given. As I said, there are only three strategies. How many charts does it take to become familiar with just one of them? Certainly not many thousands. A year's worth is enough, and that's only 250. Even if one examines every single chart, with replay that can be done in three months.

Of course one can examine thousands of charts if he wants to do so, but he would most likely have to use a computer program to do it, which would be a waste of time since he wouldn't be looking at the charts themselves, something that he will have to do if and when he begins trading. If successful backtesting were so easy to achieve via computer, there wouldn't be so very many failing traders.

Res ipsa loquitur.

 

[tradingjournals]

Quote from toolazy:

so how will you derive edge? every single anomaly is corrected sooner or later. bigger anomaly around longer, more people will notice and greedier and more violent will be reversal in fortune of this anomaly. On top of that you play against corrupt insiders which is costing you long term.

you have to learn from ichiban guy. he gets short and talk market down and he takes profit.

One day he may realize that more than 80% have to lose for 20% to win. If everyone could win who is going to pay? people who win start with the correct assumptions. And the first is that trading is zero sum game (without commission and spread). One person (people say she is a woman) indicated that one has to think of trading as a river. There is enough for all, and added that she gets paid for her time in the market. The people who lose also spend time, and they drown in the river because someone has to down for someone to win and for the broker and market maker to earn. This is not about time spend and hard working and all the empty assumptions: It is about risk in a zero sum game.